Creating Multiplayer Ubiquitous Games Using an Adaptive Narration Model Based on a User's Model

The paper proposes an adaptive narration model for Multiplayer Ubiquitous Games (MUGs), games that blend real-world and virtual-world interaction. Its central thesis is that quests can be selected and tailored to individual players by modeling each player's profile, rather than relying on the pre-scripted quest structures typical of MMORPGs. The framework draws on personality theory, specifically the Five-Factor Model and Bartle's player types, to initialize player profiles, and it treats those profiles as dynamic. A three-part feedback architecture connecting the real world, the virtual world, and the player model provides the structural basis for the approach.

The mechanics of the model centre on quest selection and instantiation. Profiles are initialized through the Five-Factor Model and updated through exponential smoothing as players make choices, using weighted vector equations. Quests fall into four classes (Destruction, Exploration, Social Interaction, and Creation), with the class chosen according to the distance between the quest and the player's profile. Each selected quest is then instantiated using location and social data. The authors illustrate this with the "Bob and Alice Game," a geo-localized, one-day-turn scenario, and classify quests as basic or complex.

The evidence presented is illustrative rather than empirical. The paper offers a conceptual framework and worked examples, and its test case, a student-developed game called Nippon Salary Racing Championship, is described without quantitative results. The authors explicitly acknowledge that coherence between the adaptive quest scheme and the underlying narrative remains an unresolved problem. The work is therefore best understood as an initial, experimental framework, with systematic evaluation identified as future work.